Anthropic has launched Claude Science, a dedicated AI workbench designed to transform how scientists conduct research. Announced on June 30, the platform integrates over 60 scientific tools and databases — including UniProt, PDB, Ensembl, and ChEMBL — into a single environment where researchers can analyze literature, execute multi-step analyses, and generate publication-ready figures and manuscripts.

Unlike general-purpose coding assistants, Claude Science acts as a coordinating agent that can spin up specialist sub-agents for genomics, proteomics, structural biology, and cheminformatics. It natively renders 3D protein structures, genome browser tracks, and chemical structures, with every output carrying a full auditable history of how it was produced.

The platform manages compute resources across local machines, HPC clusters, and on-demand GPU providers like Modal. A built-in reviewer agent inspects citations, calculations, and figures for errors, self-correcting as it goes. Researchers can fork sessions to compare approaches without losing their original thread.

Early beta users reported dramatic speedups. Manifold Bio used Claude Science to nominate drug targets end-to-end. A neuroscientist at the Allen Institute cut the time to write 100-page scientific reviews from two years to weeks. UCSF researchers completed germline genetic analyses in roughly one-tenth the previous time.

Claude Science is available in beta on macOS and Linux for Pro, Max, Team, and Enterprise users. Anthropic is also funding up to 50 AI for Science projects with $30,000 in credits each, with applications open through July 15, 2026.

Sources